以時間序列將專利文件視覺化的研究

Abstract

隨著知識經濟時代的來臨，智慧財產權倍受重視，專利的申請件數是以驚人的速度在增加，但以人工進行專利分析是件耗時耗力的工作，如何快速地從專利文獻中獲得有用的情報，成為現今相當重要的議題。

本論文主要在提出一套系統，輔佐使用者進行專利分析，呈現特定領域的技術與經營脈絡。首先，使用文字探勘的方法，擷取出專利文件中的重要概念，再採用統計檢定的方法，檢定詞與時間的關係，定義我們所謂的趨勢，最後將分析結果以視覺化的界面呈現給使用者。

With the coming of economic-based knowledge, we pay much attention to intellectual property (IP) content and the amount of patent documents increases quickly. Patent documents are the repository of how technology advances and, more importantly, show how language supports the change.[2] However, increasing patent documents makes the reading complicate and requires advanced information technology to assist the investigation of patents. In this paper, we propose a patent trend analysis system, which combines text mining and statistic test methods. We want to show the technology trend in a specific field. In the patent trend analysis system, firstly, we make definitions about concepts and apply text mining to extract important concepts from patents. Second, we use statistic test to check whether the retrieved concepts are significant in a specific time interval. Finally, we visualize the analysis result.